RESEARCH AND PRACTICE
Economies of Scale in the Production of Public Health Services: An Analysis of Local Health Districts in Florida Patrick M. Bernet, PhD, and Simone Singh, PhD
In its 2012 report, “For the Public’s Health: Investing in a Healthier Future,” the Institute of Medicine called for the development of measures to capture information on “key elements of public health delivery, including program implementation costs.”1(p7) At the same time, recent public health funding reductions put local health districts (LHDs) under increased pressure to get more done with limited resources.2,3 Borrowing from the adage, “you can’t manage what you can’t measure,” an important first step in improving the use of public health resources is to understand what it costs to provide services today. Although a growing body of research has investigated the costs of providing public health services, the findings vary widely, and precise estimates are not yet available.4---8 One of the challenges in assessing public health activities is a lack of common definitions.9 Even common services, such as well inspections, are done by staff with different training, testing for different chemicals, applying different minimum acceptable standards, and using different methods. With such inconsistencies, it is difficult to compare or benchmark costs.1 Theories of economies of scale10,11 suggest that the cost per unit decreases as more units are produced. First, fixed costs can be spread over more units, reducing per unit costs. Higher volumes also permit greater specialization of staff. For example, a small rural LHD may not have adequate patient volume for a full-time prenatal nurse and a full-time health educator. Using 1 person for both services requires LHDs to pay that staff member for the higher job classification, even when the staff member performs duties of the less-skilled, lower paying position. In this manner, cost per unit can be higher at lower volumes. Regionalization has been considered a possible path to greater efficiency for LHDs.4,5,12,13 If several smaller LHDs combined efforts, cost per service would decrease, freeing up scarce resources to expand public health services in other areas.
Objectives. We examined the existence and the extent of scale and scope economies in the delivery of public health services. We also tested the strength of agency, population, and community characteristics that moderate scale and scope economies. Methods. We collected service count and cost data for all Florida local health districts for 2008 and 2010, complemented with data on agency, population, and community characteristics. Using translog cost functions, we built models of operating efficiencies for 5 core public health activities: communicable disease surveillance, chronic disease prevention, food hygiene, on-site sewage treatment, and vital records. Results. Economies of scale were found in most activities, with cost per unit decreasing as volume increased. The models did not, however, identify meaningful economies of scope. Conclusions. Consolidation or regionalization might lower cost per unit for select public health activities. This could free up resources for use in other areas, further improving the public’s health. (Am J Public Health. 2015;105:S260–S267. doi:10.2105/AJPH.2014.302350)
Theories of economies of scope10 propose that the cost per unit decreases as more different services are produced. In effect, producing several different models of an automobile in a single factory makes the production of each model more efficient. Similarly, in public health service delivery, environmental health activities (e.g., vector control) may not employ epidemiologists. However, if an LHD’s population-based health activities have epidemiological investigators available to answer quick questions, vector control might be more productive. Regionalization scenarios might include keeping existing LHDs in place, but having each specialize in a subset of activities. So 1 LHD might provide all clinical prevention, medical treatment, and specialty care services to both LHD populations, and the other LHD would do likewise for population, regulatory, and environmental activities. A number of studies have looked at the costs of producing services for individual public health programs, such as influenza vaccination,14 federally qualified health centers clinical operations,15 or diabetes education.16 However, such studies are narrowly focused on
S260 | Research and Practice | Peer Reviewed | Bernet and Singh
a single service or a limited geography, such as a study of HIV programs that specifically examines high-risk urban women in 3 US cities.17 Although such studies are important, they are not easily generalizable. On a broader scale, a handful of studies have investigated the relationship between LHD performance and scale and scope of services.4---6 Even then, few have employed the detailed operating statistics on service counts needed to accurately estimate scale and scope economies. Such precision is a vital building block in our understanding of the costs of public health services. Unlike most states, LHDs in the state of Florida use a standard method of counting services, customers, costs, and revenues for each category of services provided. Using data from Florida, we provided an explanatory analysis of the existence, extent, and moderators of economies of scale and scope in delivering select core public health activities.
METHODS Data for our study came from 3 sources: the Florida Department of Health (FDOH), the National Association of County and City Health
American Journal of Public Health | Supplement 2, 2015, Vol 105, No. S2
RESEARCH AND PRACTICE
Officials (NACCHO), and the Area Resource File (ARF). The FDOH requires each LHD in the state to use a standard method of counting services, customers, costs, and revenues for each public health activity performed. Currently, Florida LHDs engage in up to 50 different public health activities, including core services such as inspections, communicable disease control and health promotion, as well as clinical services (e.g., immunizations and medical care).13 Florida LHDs report this information to the FDOH, which reviews the data for accuracy. The large number of service categories helps reduce service intensity variation that could make cost studies more difficult. For example, rather than lump all water-related services under a single heading, Florida has separate counts for public water, private water, pools, beach monitoring, individual sewage, public sewage, and water pollution. With so many categories, there is great homogeneity within each category, which means less variation in resource use. We obtained data for all 67 Florida LHDs for 2008 and 2010 for 5 core public health activities: communicable disease surveillance and investigation, chronic disease prevention, food hygiene services, on-site sewage treatment and disposal services, and vital records services (Table 1). We chose to study these services because they represent a diverse set of core public health activities provided by all LHDs in Florida. Together, the 5 services chosen consume approximately 17% of the average LHD’s expenditures. Categories composed of a relatively consistent mix of services were preferred to eliminate possible confounding factors. For example, beach inspection was not used because the level of service varies based on geography of beaches, industry, and level of tourist activity; different LHDs might do the same number of beach inspection services, but the mix of services might vary greatly. LHD service and cost information were supplemented with agency organizational, financial, and operational characteristics from the surveys of LHDs, which were conducted in 2008 and 2010 by the NACCHO. All 67 Florida LHDs participated in both the 2008 and 2010 Profile Studies.18 Additional county-level information on population and community characteristics came from the ARF.
TABLE 1—Descriptive Statistics for Study Variables: Florida Department of Health, the National Association of County and City Health Officials, and the Area Resource File, Florida, 2008 and 2010 Variable
Mean (SD)
Minimum–Maximum
183 407 (312 249) 2 046 (4 139)
0–2 152 738 0–23 593
225 499 (374 354)
887–2 612 716
Communicable disease surveillance and investigation Expenditures, $ Services, no. Chronic disease prevention Expenditures, $ Services, no.
880 (2 279)
0–19 477
Food hygiene Expenditures, $
106 045 (154 166)
125–737 441
1 039 (1 546)
4–9 365
373 577 (337 513)
39 103–1 832 989
3 368 (3 832)
196–21 094
180 409 (284 042)
42–1 779 575
27 917 (38 766)
448–179 091
Clinical preventive services, %
89 (16)
0–100
Medical care services, % Specialty services, %
63 (17) 39 (26)
0–88 0–100
Population-based services, %
59 (17)
0–92
Regulatory and licensing, %
62 (19)
0–100
Environmental health, %
36 (27)
0–100
Services, no. On-site sewage treatment and disposal Expenditures, $ Services, no. Vital records Expenditures, $ Services, no. Scope of services offered
Public health agency characteristics Revenue from local sources, %
8 (7)
0–33
Revenue from state sources, %
36 (11)
4–69
Revenue from federal sources, % Revenue from clinical services, %
15 (8) 24 (14)
3–44 0–71
318 (503)
10–3346
Population and community characteristics 1000 persons per square mile Population non-White, %
20 (10)
6–64
Persons with college education, %
13 (6)
4–27
Persons aged ‡ 65 y, % Income per capita, $ Persons insured, %
Average characteristics of LHDs are displayed in Table 1, along with variance and high-low ranges that demonstrate the diversity in the sociodemographic and geographic characteristics of Florida counties. FDOH information showed, for example, that the average LHD produced 2046 communicable disease surveillance and investigation services at an average total cost of $183 407. Scope of
Supplement 2, 2015, Vol 105, No. S2 | American Journal of Public Health
18 (6)
8–43
32 653 (10 081)
18 421–62 559
24 (4)
16–37
services data from NACCHO showed most LHDs (89%) provided clinical preventive services, but only 36% performed any environmental health service. Community characteristics from ARF showed that the average LHD covered 318 square miles; 24% of its patients were uninsured, 20% were non-White, and there were more people older than 65 years than there were college graduates.
Bernet and Singh | Peer Reviewed | Research and Practice | S261
RESEARCH AND PRACTICE
Measures of Cost of Service Provision The dependent variable of interest was the total cost of providing each of the following 5 core public health services: communicable disease surveillance and investigation, chronic disease prevention, food hygiene services, onsite sewage treatment and disposal services, and vital records services (Table 2). Information on expenditures incurred was obtained from the FDOH. The previously detailed service categorizations are complemented with the FDOH’s use of consistent accounting policies and procedures in all LHDs, which makes spending, staffing, and volume reporting comparable across the state. FDOH includes both direct and indirect cost components to assure as complete a picture as possible.19,20 Because total costs were highly skewed, we used the natural logarithm of total costs in all analyses. In addition to total costs, we also computed per unit costs for each of the 5 activities by dividing total expenditures into the number of services provided. Total costs and cost per service varied widely across LHDs (Table 2). The median cost per service in food hygiene, for instance, was $104 in 2010. The 25% of LHDs with the lowest costs per service, however, spent less than $74 per service, whereas the most costly 25% spent more than $133, or 0.8 times more. Similar patterns were seen for the other core public health activities. Chronic disease prevention services were the most costly, with median per service costs of $470. In contrast, communicable disease surveillance and investigation, exhibited the largest variation in costs, with the most costly 25% of LHDs spending 5.8 times more than the least costly 25%.
Measures of Economies of Scale and Scope Key independent variables of interest were indicators of economies of scale and scope. We measured economies of scale using information from FDOH on the number of services performed in each of the 5 categories of public health activities examined. We measured economies of scope using information from the NACCHO Profile Studies, which assessed if a service category was performed, regardless of the volume of service within that category. For example, the NACCHO service category diabetes screening services would be considered performed by the LHD whether they provided 1000 such services, or just one. In other words, NACCHO service categories measured whether a service was performed, not how much or how often it was performed. NACCHO collects data on more than 40 different service categories commonly provided by LHDs. To estimate the scope of services performed, we collapsed NACCHO service categories into 6 major classes that have been employed in previous studies21: clinical preventive services, medical care services, specialty services, populationbased services, regulatory and licensing services, and environmental health services. We then measured the breadth of coverage within each class as the proportion of all service categories that the LHD performed in-house. For example, an LHD that performed 9 of 11 clinical preventive service categories in-house was assigned a score of 82% in that category.
Analysis We used a translog cost model to estimate how economies of scale and scope were
associated with LHDs’ costs of providing select public health services; we controlled for agency, population, and community characteristics (Table 1). This is a standard approach that has been employed in numerous public health studies.4---6,10,22---24 The model expressed the natural logarithm of the total costs of a service line as a function of the number of service provided (service_count), the squared number of service provided (service_count2), the scope of services provided, and a set of agency, population, and community characteristics (Equation 1).21,25 Of the 134 LHD-year observations, we dropped several observations with $0 spending to accommodate the log transformation, which resulted in a sample size of 115 (111 for communicable disease surveillance and investigation). To test for economies of scale, our model included the number of services and the square of the number of services provided in each service line. We frequently used quadratic functions in applied economics to capture increasing or decreasing marginal effects.10 In our model, the presence of economies of scale was indicated by a positive coefficient on the number of services provided and the diminishing returns to scale registers as a negative coefficient on the squared number of service provided. To model the economies of scope, we used separate measures for each of the 6 categories of service to indicate the proportion of services in each category provided directly by the LDH. Using cross-sectional, ordinary least-squares regressions, we estimated separate models for each of the 5 activities (Equation 1). Breusch-Pagan tests indicated that heteroskedasticity was not a concern.
TABLE 2—Total Cost and Cost per Service for 5 Select Public Health Services: Florida Department of Health, the National Association of County and City Health Officials, and the Area Resource File, Florida, 2008 and 2010 Cost per Service Public Health Activity
Total Cost, Mean (SD)
Median
Communicable disease surveillance and investigation
$177 640 ($296 456)
$128
Chronic disease prevention
$187 911 ($295 105)
$470
Food hygiene
$107 910 ($151 103)
$104
$74
$133
0.8
On-site sewage treatment and disposal
$329 423 ($291 648)
$141
$101
$209
1.1
Vital records
$181 276 ($274 017)
S262 | Research and Practice | Peer Reviewed | Bernet and Singh
$6.37
25th Percentile
75th Percentile
75th: 25th Percentile Ratio
$74
$503
5.8
$290
$1 446
4.0
$5.18
$9.73
0.9
American Journal of Public Health | Supplement 2, 2015, Vol 105, No. S2
RESEARCH AND PRACTICE
ð1Þ Translog regression model : LogðTotal CostÞ ¼a þ b1 service count þ b2 service count2 þ b3 Scope of categories of service þ b4 Community þe
RESULTS Multivariate parameter estimates showed that the cost of public health activities varied
significantly with the volume of services provided (Table 3). By contrast, the scope of services provided was not associated with meaningful changes in LHDs’ costs of providing any of the activities examined in our study. We examined scale effects first; there was evidence of economies of scale for communicable disease surveillance and investigation, food hygiene, on-site sewage treatment and disposal, and vital records activities. The positive coefficient on the service count variable indicated that total costs increased at higher volumes. By contrast, the negative coefficient on the squared services term indicated that
those costs increased at a decreasing rate. In other words, the cost per unit decreased as the volume of services provided increased. The strength of the scale effects is apparent in Figures 1 to 3, which plot costs and service volumes for 3 of the 5 services studied. Both types of graphs provide evidence of economies of scale. The graphs for total costs on the left show that total costs rose with increases in the volume of services provided, yet the rate at which costs increased slows for larger volumes. The graphs for cost per service on the right show that as volumes increased per unit, costs decreased. Although the general graph shapes
TABLE 3—Regression Estimates of Factors Associated With Logged Total Costs of Providing Select Public Health Services: Florida Department of Health, the National Association of County and City Health Officials, and the Area Resource File, Florida, 2008 and 2010 Parameter Estimates Scale of Services Offered Servicesa Services squaredb
Communicable Disease Surveillance and Investigation (n = 111) 0.48** –0.19**
Chronic Disease Prevention (n = 115) 0.23 –0.09
Food Hygiene (n = 115)
On-site Sewage Treatment and Disposal (n = 111)
Vital Records (n = 111)
1.95** –0.31**
3.50** –0.013**
0.074** –0.00034** –1.35
Scope of services offered Clinical preventive services, %
–0.76
0.95
0.59
–0.07
Medical care services, %
–0.67
0.32
–0.35
–0.15
0.12
Specialty services, %
0.13
–0.30
–0.22
0.22
–0.25
Population-based services, %
1.46
1.53*
0.14
–0.31
–0.78
Regulatory and licensing, %
0.87
0.34
0.52
0.10
0.67
Environmental health, % Public health agency characteristics
0.04
–0.82
–0.31
–0.10
0.46
Revenue from local sources, %
–3.47
–1.56
2.39
–0.20
–0.15
Revenue from state sources, %
–3.23
2.10
–0.47
–0.79
–2.70
Revenue from federal sources, %
–3.16
2.88
1.55
–1.37*
–4.91
Revenue from clinical services, %
–3.90*
1.12
–0.46
–0.79
–0.74
1000 persons per square mile
–0.02
0.88**
0.26
0.08
–0.05
Population non-White, % Persons with college education, %
–0.01 11.31*
0.01 2.27
0.02* 4.79
0.00 1.78
–0.01 8.34*
Persons aged ‡ 65 y, %
0.85
–0.24
2.39
–1.02
–0.96
Income per capita (log)
1.26
1.20
0.30
0.10
–0.68
Percent uninsured, %
0.00
0.02
0.02
–0.02
0.04
Year 2008c
0.22
0.72**
0.14
0.16
–0.04
Population and community characteristics
Model Constant Adjusted R2
–2.14
–5.62
3.39
11.71**
18.10*
0.68
0.56
0.84
0.86
0.71
a
Services were expressed in terms of 1000 units. Services squared were expressed in terms of 1 000 000 (1000 x 1000) units provided. Reference year 2010. *P < .05; **P < .01. b c
Supplement 2, 2015, Vol 105, No. S2 | American Journal of Public Health
Bernet and Singh | Peer Reviewed | Research and Practice | S263
RESEARCH AND PRACTICE
a
b 700
1 200 000
600
1 000 000
500
Cost, $
Cost, $
800 000 600 000
400 300
400 000
200
200 000
100
0
0
5 000
10 000
15 000
20 000
25 000
0
0
5 000
10 000
15 000
20 000
25 000
Number of Services
Number of Services
FIGURE 1—Expense and number of services in local health departments for communicable disease surveillance and investigation by (a) total cost and (b) cost per service: Florida Department of Health, the National Association of County and City Health Officials, and the Area Resource File, Florida, 2008 and 2010.
are similar, subtle differences might point to differences in production processes. For example, the cost per service curves for communicable disease surveillance and investigation bends sharply at a relatively low volume, showing fairly flat cost per service beyond that threshold level. This indicated that LHDs might benefit from economies of scale up to the threshold level, but that they might not realize meaningful gains beyond. By contrast, the more gradual slope for per service costs of onsite sewage treatment and disposal indicates that scale economies appeared to be present across the entire range of service volume. Economies of scale can also be thought of in per unit terms, so we duplicated study regressions to estimate cost per service. These alternate models again found scale economies for communicable disease surveillance and investigation, on-site sewage inspection, vital records, and chronic disease prevention services, all of which had negative coefficients on volume, which indicated lower cost per unit at higher volumes. These per unit cost models had the advantage of being more comparable across services, because they were less subject to the sheer impact of population size that was demonstrated in earlier studies.4---6 We also found another perspective on how efficiency considers cost per client. We re-ran study models to estimate total costs based on number of clients rather than service volume. All services demonstrated economies of scale
in per client costs, which was likely caused by the fairly uniform services-per-client intensity across all LHDs. In addition to the fine-grained service classifications noted earlier, Florida has rigorous procedures in place to ensure consistent coding and counting of services, costs, and clients. Looking at scope effects, we found no evidence of meaningful economies of scope for any of the 5 public health activities examined in this study (Table 3). With 1 exception, all coefficients on indicators of the scope of services provided were not significant. Similarly, public health agency, population, and community characteristics were largely unrelated to the cost of service provision (Table 3).
DISCUSSION LHDs varied considerably in size, range of services offered, and the demographic characteristics of the population served. The cost of providing public health services also varied widely. Our findings regarding the cost of delivering services, and the agency, population, and community characteristics that influenced those costs, provided valuable information for agency decisions regarding the scale and scope of public health operations. Our study was among the first to systematically search for economies of scale and scope in the delivery of public health activities. Our findings indicated that economies of scale arose
S264 | Research and Practice | Peer Reviewed | Bernet and Singh
in the production of many core public health activities. This was hardly surprising. Many public health services require a significant level of fixed capacity or staffing, which results in high fixed costs. The recordkeeping infrastructure behind vital records activities, for instance, requires considerable investment in information technologies. Spreading these fixed costs across more services could help health districts reduce per service costs. The presence of economies of scale in the delivery of public health services had important implications for the organization of public health at the local level. Increasing the number of services provided in certain categories might allow LHDs to reduce the cost per unit of service, thus freeing up resources for use in other areas. Evidence of economies of scale could thus inform discussions of local health district consolidation, regionalization, and the provision of shared services. In situations in which an LHD simply does not need the number of services necessary to achieve economies of scale, specialization might help. If 1 LHD provides the service to a number of other LHDs, all would benefit from lower per unit costs. However, focusing attention on a larger production process for 1 service takes attention away from other services, for which local needs are also too low to achieve economies. Rather than each LHD trying to do every service alone, a strategy of specialization and sharing would help LHDs get more for the same cost.
American Journal of Public Health | Supplement 2, 2015, Vol 105, No. S2
RESEARCH AND PRACTICE
a
b 800 000
250
700 000
200
500 000
Cost, $
Cost, $
600 000
400 000
100
300 000 200 000
50
100 000 0
150
0
0
1 000 2 000 3 000 4 000 5 000 6 000 7 000 8 000 9 000 10 000
0
1 000 2 000 3 000 4 000 5 000 6 000 7 000 8 000 9 000 10 000
Number of Services
Number of Services
FIGURE 2—Expense and number of services in local health departments for food hygiene by (a) total cost and (b) cost per service: Florida Department of Health, the National Association of County and City Health Officials, and the Area Resource File, Florida, 2008 and 2010.
An individual LHD might, for example, provide food hygiene services for the entire region, but discontinue the direct provision of on-site sewage inspection in favor of contracted services from another LHD in that region. This broader perspective allows LHDs, as a group, to achieve the economies of scale that none could reach individually. At that individual level, however, cost savings are just 1 factor in strategic decisions that must also consider
strong desires for local governance, potential legal barriers to sharing services, management practices and staff training needed to enable large-scale production, special circumstances of small and rural LHDs, concerns that savings in 1 area must be returned rather than invested in other services,22 and the degree of similarity of adjacent LHD populations that could moderate efficiency levels actually realized. If these factors align, LHDs have a chance to provide more
a
b 2 000 000
450.00
1 800 000
400.00
1 600 000
350.00
1 400 000
300.00
1 200 000
Cost, $
Cost, $
services, effectively improving population health for no additional cost. Findings of economies of scale, however, also pose challenges, especially for smaller LHDs. Higher costs per unit at lower volumes mean that smaller LHDs might not be able to engage in the same level of public health activities, even if their budgets and staffing per capita are equal to those of larger LHDs. Staffing decisions can be challenging at low
1 000 000 800 000
200.00 150.00
600 000
100.00
400 000
50.00
200 000 0
250.00
0.00 0
5 000
10 000
15 000
20 000
Number of Services
25 000
0
5 000
10 000
15 000
20 000
Number of Services
FIGURE 3—Expense and number of services in local health departments for on-site sewage treatment and disposal by (a) total cost and (b) cost per service: Florida Department of Health, the National Association of County and City Health Officials, and the Area Resource File, Florida, 2008 and 2010.
Supplement 2, 2015, Vol 105, No. S2 | American Journal of Public Health
Bernet and Singh | Peer Reviewed | Research and Practice | S265
RESEARCH AND PRACTICE
volumes. Highly specialized workers are often more efficient, but that very specialization limits the functions they can perform. This can leave LHDs with a choice between paying for highly trained workers who are overqualified (and overpaid) for many of the tasks to which they are assigned or using lower qualified workers who may be less efficient. Our findings indicated that small LHDs might thus need to rely on contracted services or partnerships with larger LHDs when attempting to reduce the cost of service provision. Unfortunately, the potential cost savings depend largely on the rate at which services could be subcontracted, and smaller LHDs might lack the bargaining power to negotiate favorable rates.26,27 The difficulties experienced by smaller LHDs might also explain the broader variation in cost per unit at lower volumes that we found (Figures 1---3), making managing smaller LHDs even more of a challenge. By contrast, the breadth of services provided by a LHD was not associated with meaningful variation in the cost of service provision. These findings suggested that, at least for the activities we examined in this study, there were neither economies nor diseconomies of scope. This was an encouraging finding in a public health setting in which many LHDs were expected to provide at least some minimum package of core services. Because the scope of services did not appear to affect the costs of performing individual public health activities, decisions regarding the range of public health services provided might be based on factors other than financial ones and instead follow strategic considerations because of legal requirements.
Limitations Any generalization of our findings must acknowledge the limitations of our study, which we performed using data from LHDs in only 1 state, Florida, and only for 2 years, 2008 and 2010. Many LHDs experienced resource cuts and higher service demand during this time,10 and future research is needed to understand the factors that drive LHDs’ costs in times of stable or growing budgets. Moreover, our cross-sectional study design did not allow inferences regarding the causality of the relationships presented. Longitudinal studies might be better able to identify possible endogeneity issues, wherein LHDs could self-select
lower volume because of higher costs. Although economic theory predicts that greater volume of services will result in lower costs, it is theoretically possible that LHDs may self-select lower volume because of higher costs. In addition, we only studied 5 services, so these services might not be generalizable to others. Exporting findings to other states would require further investigation into other LHD characteristics, such as executive training, board of health powers, and recognition that Florida is a centralized state. Finally, replicating total cost, per unit cost, and per capita models with other services could lead to a deeper understanding of the role of service volume, population size, and breadth of service coverage in the cost of delivering public health services.
Conclusions 1,2
We continued a line of research that examined the cost of providing public health services. The strength of our study was that the models presented were able to explain a significant portion of the variation in costs. This made these models a useful tool for both public health practitioners and researchers. Resource requirements could be estimated with greater precision using such models to better inform policymakers about funding requirements. An understanding of economies of scale is also relevant to discussions about consolidating the production of services and the creating of shared programs across LHDs. Funds saved as a result could be spent on other public health activities, resulting in a net improvement in the health of a population at no additional cost. j
Acknowledgments This project was partly funded by a grant from the Robert Wood Johnson Foundation (grant N016299).
Human Participant Protection This study was approved by the University of Michigan. No human participants were involved.
References 1. Committee on Public Health Strategies to Improve Health; Board on Population Health and Public Health Practices; Institute of Medicine. For the Public’s Health: Investing in a Healthier Future. Available at: https:// download.nap.edu/catalog.php?record_id=13268. Accessed March 3, 2013. 2. Leider JP, Resnick B, Kass N, et al. Budget- and priority-setting criteria at state health agencies in times of austerity: a mixed methods study. Am J Public Health. 2014;104(6):1092---1099. 3. Willard R, Shah GH, Leep C, Ku L. Impact of the 2008-2010 economic recession on local health departments. J Public Health Manag Pract. 2012;18(2):106---114. 4. Santerre RE. Jurisdiction size and local public health spending. Health Serv Res. 2009;44(6):2148---2166. 5. Mays GP, McHugh MC, Shim K, et al. Institutional and economic determinants of public health system performance. Am J Public Health. 2006;96(3):523---531. 6. Erwin PC. The performance of local health departments: a review of the literature. J Public Health Manag Pract. 2008;14(2):E9---E18. 7. Ingram RC, Bernet PM, Costich JF. Public health services and systems research: current state of finance research. J Public Health Manag Pract. 2012;18(6): 515---519. 8. Singh SR. Efficiency in public health service delivery: an analysis of clinical health services provided by local health departments in Florida. Frontiers Public Health Serv Sys Res. 2013;2(7):Article 3. 9. Honoré PA, Amy BW. Public health finance: fundamental theories, concepts, and definitions. J Public Health Manag Pract. 2007;13(2):89---92. 10. Wooldridge JM. Introductory Econometrics: A Modern Approach. 5th ed. Mason, OH: South-Western Cengage Learning; 2012.
About the Authors Patrick M. Bernet is with the College of Business, Florida Atlantic University, Boca Raton. Simone Singh is with the School of Public Health, University of Michigan, Ann Arbor. Correspondence should be sent to Patrick M. Bernet, Florida Atlantic University, College of Business, Department of Health Administration, 127 Fleming West, 777 Glades Road, Boca Raton, FL 33431 (e-mail: PBernet@ fau.edu). Reprints can be ordered at http://www.ajph.org by clicking the “Reprints” link. This article was accepted September 18, 2014.
Contributors This project was jointly conceptualized by P. M. Bernet and S. Singh. P. M. Bernet performed the background research, organized data, and led the writing of the article. S. Singh contributed to the background research,
S266 | Research and Practice | Peer Reviewed | Bernet and Singh
performed statistical analysis, and assisted in writing the technical parts of the article.
11. Getzen TE. Health Economics and Financing. 5th ed. New York, NY: Wiley; 2012. 12. Institute of Medicine. The Future of the Public’s Health in the 21st Century. Washington, DC: The National Academies Press; 2002. Available at: http://www.iom. edu/Reports/2002/The-Future-of-the-Publics-Healthin-the-21st-Century.aspx. Accessed December 12, 2012. 13. Frieden TR. A framework for public health action, the health impact pyramid. Am J Public Health. 2010; 100(4):590---595. 14. Bridges CB, Thompson W, Meltzer MI, et al. Efficacy and cost benefit of influenza vaccination of healthy working adults: a randomized controlled trial. JAMA. 2000;284:1655---1663. 15. Amico PR, Chilingerian JA, van Hasselt M. Community health center efficiency: the role of grant
American Journal of Public Health | Supplement 2, 2015, Vol 105, No. S2
RESEARCH AND PRACTICE
revenues in health center efficiency. Health Serv Res. 2014;49:666---682. 16. Keers JC, Groen H, Sluiter WJ, Bouma J, Links TP. Cost and benefits of a multidisciplinary intensive diabetes education programme. J Eval Clin Pract. 2005;11(3): 293---303. 17. Chesson HW, Greenberg JB, Hennessy M. The cost effectiveness of the WINGS intervention: a program to prevent HIV and sexually transmitted diseases among high risk urban women. BMC Infect Dis. 2002;2:24. 18. Leep CJ, Shah GH. NACCHO’s national profile of local health departments study: the premier source of data on local health departments for surveillance, research, and policymaking. J Public Health Manag Pract. 2012;18(2):186---189. 19. Division of Environmental Health. Environmental Health Coding Pamphlet. DHP-50-21 (V2). Tallahassee, FL: Florida Department of Health; 2011. 20. Department of Health. Personal Health Coding Pamphlet. DHP-50-20. Tallahassee, FL: Florida Department of Health; 2012. 21. Mays GP, Smith S. Geographic variation in public health spending: correlates and consequences. Health Serv Res. 2009;44(5 pt 2):1796---1817. 22. Bernet PM. Local public health agency funding: money begets money. J Public Health Manag Pract. 2007;13(2):188---193. 23. Erwin PC, Greene SB, Mays GP, Ricketts TC, Davis MV. The association of changes in local health department resources with changes in state-level health outcomes. Am J Public Health. 2011;101(4):609---615. 24. Reschovsky A, Zahner S, Enami K, Goitom S. Forecasting the impact of the economic recession on the financing of local public health departments in Wisconsin. Presented at: Public Health Practice Based Research Networks Meeting [webinar]: August 16, 2012. 25. Mukherjee K, Santerre RE, Zhang NJ. Explaining the efficiency of local health departments in the US: an exploratory analysis. Health Care Manag Sci. 2010; 13(4):378---387. 26. Gollust SE, Jacobson PD. Privatization of public services: organizational reform efforts in public education and public health. Am J Public Health. 2006;96(10): 1733---1739. 27. Keane C, Marx J, Ricci E. Perceived outcomes of public health privatization: a national survey of local health department directors. Milbank Q. 2001;79(1): 115---137.
Supplement 2, 2015, Vol 105, No. S2 | American Journal of Public Health
Bernet and Singh | Peer Reviewed | Research and Practice | S267
Economies of Scale in the Production of Public Health Services: An Analysis of Local Health Districts in Florida Patrick M. Bernet, PhD, and Simone Singh, PhD
In its 2012 report, “For the Public’s Health: Investing in a Healthier Future,” the Institute of Medicine called for the development of measures to capture information on “key elements of public health delivery, including program implementation costs.”1(p7) At the same time, recent public health funding reductions put local health districts (LHDs) under increased pressure to get more done with limited resources.2,3 Borrowing from the adage, “you can’t manage what you can’t measure,” an important first step in improving the use of public health resources is to understand what it costs to provide services today. Although a growing body of research has investigated the costs of providing public health services, the findings vary widely, and precise estimates are not yet available.4---8 One of the challenges in assessing public health activities is a lack of common definitions.9 Even common services, such as well inspections, are done by staff with different training, testing for different chemicals, applying different minimum acceptable standards, and using different methods. With such inconsistencies, it is difficult to compare or benchmark costs.1 Theories of economies of scale10,11 suggest that the cost per unit decreases as more units are produced. First, fixed costs can be spread over more units, reducing per unit costs. Higher volumes also permit greater specialization of staff. For example, a small rural LHD may not have adequate patient volume for a full-time prenatal nurse and a full-time health educator. Using 1 person for both services requires LHDs to pay that staff member for the higher job classification, even when the staff member performs duties of the less-skilled, lower paying position. In this manner, cost per unit can be higher at lower volumes. Regionalization has been considered a possible path to greater efficiency for LHDs.4,5,12,13 If several smaller LHDs combined efforts, cost per service would decrease, freeing up scarce resources to expand public health services in other areas.
Objectives. We examined the existence and the extent of scale and scope economies in the delivery of public health services. We also tested the strength of agency, population, and community characteristics that moderate scale and scope economies. Methods. We collected service count and cost data for all Florida local health districts for 2008 and 2010, complemented with data on agency, population, and community characteristics. Using translog cost functions, we built models of operating efficiencies for 5 core public health activities: communicable disease surveillance, chronic disease prevention, food hygiene, on-site sewage treatment, and vital records. Results. Economies of scale were found in most activities, with cost per unit decreasing as volume increased. The models did not, however, identify meaningful economies of scope. Conclusions. Consolidation or regionalization might lower cost per unit for select public health activities. This could free up resources for use in other areas, further improving the public’s health. (Am J Public Health. 2015;105:S260–S267. doi:10.2105/AJPH.2014.302350)
Theories of economies of scope10 propose that the cost per unit decreases as more different services are produced. In effect, producing several different models of an automobile in a single factory makes the production of each model more efficient. Similarly, in public health service delivery, environmental health activities (e.g., vector control) may not employ epidemiologists. However, if an LHD’s population-based health activities have epidemiological investigators available to answer quick questions, vector control might be more productive. Regionalization scenarios might include keeping existing LHDs in place, but having each specialize in a subset of activities. So 1 LHD might provide all clinical prevention, medical treatment, and specialty care services to both LHD populations, and the other LHD would do likewise for population, regulatory, and environmental activities. A number of studies have looked at the costs of producing services for individual public health programs, such as influenza vaccination,14 federally qualified health centers clinical operations,15 or diabetes education.16 However, such studies are narrowly focused on
S260 | Research and Practice | Peer Reviewed | Bernet and Singh
a single service or a limited geography, such as a study of HIV programs that specifically examines high-risk urban women in 3 US cities.17 Although such studies are important, they are not easily generalizable. On a broader scale, a handful of studies have investigated the relationship between LHD performance and scale and scope of services.4---6 Even then, few have employed the detailed operating statistics on service counts needed to accurately estimate scale and scope economies. Such precision is a vital building block in our understanding of the costs of public health services. Unlike most states, LHDs in the state of Florida use a standard method of counting services, customers, costs, and revenues for each category of services provided. Using data from Florida, we provided an explanatory analysis of the existence, extent, and moderators of economies of scale and scope in delivering select core public health activities.
METHODS Data for our study came from 3 sources: the Florida Department of Health (FDOH), the National Association of County and City Health
American Journal of Public Health | Supplement 2, 2015, Vol 105, No. S2
RESEARCH AND PRACTICE
Officials (NACCHO), and the Area Resource File (ARF). The FDOH requires each LHD in the state to use a standard method of counting services, customers, costs, and revenues for each public health activity performed. Currently, Florida LHDs engage in up to 50 different public health activities, including core services such as inspections, communicable disease control and health promotion, as well as clinical services (e.g., immunizations and medical care).13 Florida LHDs report this information to the FDOH, which reviews the data for accuracy. The large number of service categories helps reduce service intensity variation that could make cost studies more difficult. For example, rather than lump all water-related services under a single heading, Florida has separate counts for public water, private water, pools, beach monitoring, individual sewage, public sewage, and water pollution. With so many categories, there is great homogeneity within each category, which means less variation in resource use. We obtained data for all 67 Florida LHDs for 2008 and 2010 for 5 core public health activities: communicable disease surveillance and investigation, chronic disease prevention, food hygiene services, on-site sewage treatment and disposal services, and vital records services (Table 1). We chose to study these services because they represent a diverse set of core public health activities provided by all LHDs in Florida. Together, the 5 services chosen consume approximately 17% of the average LHD’s expenditures. Categories composed of a relatively consistent mix of services were preferred to eliminate possible confounding factors. For example, beach inspection was not used because the level of service varies based on geography of beaches, industry, and level of tourist activity; different LHDs might do the same number of beach inspection services, but the mix of services might vary greatly. LHD service and cost information were supplemented with agency organizational, financial, and operational characteristics from the surveys of LHDs, which were conducted in 2008 and 2010 by the NACCHO. All 67 Florida LHDs participated in both the 2008 and 2010 Profile Studies.18 Additional county-level information on population and community characteristics came from the ARF.
TABLE 1—Descriptive Statistics for Study Variables: Florida Department of Health, the National Association of County and City Health Officials, and the Area Resource File, Florida, 2008 and 2010 Variable
Mean (SD)
Minimum–Maximum
183 407 (312 249) 2 046 (4 139)
0–2 152 738 0–23 593
225 499 (374 354)
887–2 612 716
Communicable disease surveillance and investigation Expenditures, $ Services, no. Chronic disease prevention Expenditures, $ Services, no.
880 (2 279)
0–19 477
Food hygiene Expenditures, $
106 045 (154 166)
125–737 441
1 039 (1 546)
4–9 365
373 577 (337 513)
39 103–1 832 989
3 368 (3 832)
196–21 094
180 409 (284 042)
42–1 779 575
27 917 (38 766)
448–179 091
Clinical preventive services, %
89 (16)
0–100
Medical care services, % Specialty services, %
63 (17) 39 (26)
0–88 0–100
Population-based services, %
59 (17)
0–92
Regulatory and licensing, %
62 (19)
0–100
Environmental health, %
36 (27)
0–100
Services, no. On-site sewage treatment and disposal Expenditures, $ Services, no. Vital records Expenditures, $ Services, no. Scope of services offered
Public health agency characteristics Revenue from local sources, %
8 (7)
0–33
Revenue from state sources, %
36 (11)
4–69
Revenue from federal sources, % Revenue from clinical services, %
15 (8) 24 (14)
3–44 0–71
318 (503)
10–3346
Population and community characteristics 1000 persons per square mile Population non-White, %
20 (10)
6–64
Persons with college education, %
13 (6)
4–27
Persons aged ‡ 65 y, % Income per capita, $ Persons insured, %
Average characteristics of LHDs are displayed in Table 1, along with variance and high-low ranges that demonstrate the diversity in the sociodemographic and geographic characteristics of Florida counties. FDOH information showed, for example, that the average LHD produced 2046 communicable disease surveillance and investigation services at an average total cost of $183 407. Scope of
Supplement 2, 2015, Vol 105, No. S2 | American Journal of Public Health
18 (6)
8–43
32 653 (10 081)
18 421–62 559
24 (4)
16–37
services data from NACCHO showed most LHDs (89%) provided clinical preventive services, but only 36% performed any environmental health service. Community characteristics from ARF showed that the average LHD covered 318 square miles; 24% of its patients were uninsured, 20% were non-White, and there were more people older than 65 years than there were college graduates.
Bernet and Singh | Peer Reviewed | Research and Practice | S261
RESEARCH AND PRACTICE
Measures of Cost of Service Provision The dependent variable of interest was the total cost of providing each of the following 5 core public health services: communicable disease surveillance and investigation, chronic disease prevention, food hygiene services, onsite sewage treatment and disposal services, and vital records services (Table 2). Information on expenditures incurred was obtained from the FDOH. The previously detailed service categorizations are complemented with the FDOH’s use of consistent accounting policies and procedures in all LHDs, which makes spending, staffing, and volume reporting comparable across the state. FDOH includes both direct and indirect cost components to assure as complete a picture as possible.19,20 Because total costs were highly skewed, we used the natural logarithm of total costs in all analyses. In addition to total costs, we also computed per unit costs for each of the 5 activities by dividing total expenditures into the number of services provided. Total costs and cost per service varied widely across LHDs (Table 2). The median cost per service in food hygiene, for instance, was $104 in 2010. The 25% of LHDs with the lowest costs per service, however, spent less than $74 per service, whereas the most costly 25% spent more than $133, or 0.8 times more. Similar patterns were seen for the other core public health activities. Chronic disease prevention services were the most costly, with median per service costs of $470. In contrast, communicable disease surveillance and investigation, exhibited the largest variation in costs, with the most costly 25% of LHDs spending 5.8 times more than the least costly 25%.
Measures of Economies of Scale and Scope Key independent variables of interest were indicators of economies of scale and scope. We measured economies of scale using information from FDOH on the number of services performed in each of the 5 categories of public health activities examined. We measured economies of scope using information from the NACCHO Profile Studies, which assessed if a service category was performed, regardless of the volume of service within that category. For example, the NACCHO service category diabetes screening services would be considered performed by the LHD whether they provided 1000 such services, or just one. In other words, NACCHO service categories measured whether a service was performed, not how much or how often it was performed. NACCHO collects data on more than 40 different service categories commonly provided by LHDs. To estimate the scope of services performed, we collapsed NACCHO service categories into 6 major classes that have been employed in previous studies21: clinical preventive services, medical care services, specialty services, populationbased services, regulatory and licensing services, and environmental health services. We then measured the breadth of coverage within each class as the proportion of all service categories that the LHD performed in-house. For example, an LHD that performed 9 of 11 clinical preventive service categories in-house was assigned a score of 82% in that category.
Analysis We used a translog cost model to estimate how economies of scale and scope were
associated with LHDs’ costs of providing select public health services; we controlled for agency, population, and community characteristics (Table 1). This is a standard approach that has been employed in numerous public health studies.4---6,10,22---24 The model expressed the natural logarithm of the total costs of a service line as a function of the number of service provided (service_count), the squared number of service provided (service_count2), the scope of services provided, and a set of agency, population, and community characteristics (Equation 1).21,25 Of the 134 LHD-year observations, we dropped several observations with $0 spending to accommodate the log transformation, which resulted in a sample size of 115 (111 for communicable disease surveillance and investigation). To test for economies of scale, our model included the number of services and the square of the number of services provided in each service line. We frequently used quadratic functions in applied economics to capture increasing or decreasing marginal effects.10 In our model, the presence of economies of scale was indicated by a positive coefficient on the number of services provided and the diminishing returns to scale registers as a negative coefficient on the squared number of service provided. To model the economies of scope, we used separate measures for each of the 6 categories of service to indicate the proportion of services in each category provided directly by the LDH. Using cross-sectional, ordinary least-squares regressions, we estimated separate models for each of the 5 activities (Equation 1). Breusch-Pagan tests indicated that heteroskedasticity was not a concern.
TABLE 2—Total Cost and Cost per Service for 5 Select Public Health Services: Florida Department of Health, the National Association of County and City Health Officials, and the Area Resource File, Florida, 2008 and 2010 Cost per Service Public Health Activity
Total Cost, Mean (SD)
Median
Communicable disease surveillance and investigation
$177 640 ($296 456)
$128
Chronic disease prevention
$187 911 ($295 105)
$470
Food hygiene
$107 910 ($151 103)
$104
$74
$133
0.8
On-site sewage treatment and disposal
$329 423 ($291 648)
$141
$101
$209
1.1
Vital records
$181 276 ($274 017)
S262 | Research and Practice | Peer Reviewed | Bernet and Singh
$6.37
25th Percentile
75th Percentile
75th: 25th Percentile Ratio
$74
$503
5.8
$290
$1 446
4.0
$5.18
$9.73
0.9
American Journal of Public Health | Supplement 2, 2015, Vol 105, No. S2
RESEARCH AND PRACTICE
ð1Þ Translog regression model : LogðTotal CostÞ ¼a þ b1 service count þ b2 service count2 þ b3 Scope of categories of service þ b4 Community þe
RESULTS Multivariate parameter estimates showed that the cost of public health activities varied
significantly with the volume of services provided (Table 3). By contrast, the scope of services provided was not associated with meaningful changes in LHDs’ costs of providing any of the activities examined in our study. We examined scale effects first; there was evidence of economies of scale for communicable disease surveillance and investigation, food hygiene, on-site sewage treatment and disposal, and vital records activities. The positive coefficient on the service count variable indicated that total costs increased at higher volumes. By contrast, the negative coefficient on the squared services term indicated that
those costs increased at a decreasing rate. In other words, the cost per unit decreased as the volume of services provided increased. The strength of the scale effects is apparent in Figures 1 to 3, which plot costs and service volumes for 3 of the 5 services studied. Both types of graphs provide evidence of economies of scale. The graphs for total costs on the left show that total costs rose with increases in the volume of services provided, yet the rate at which costs increased slows for larger volumes. The graphs for cost per service on the right show that as volumes increased per unit, costs decreased. Although the general graph shapes
TABLE 3—Regression Estimates of Factors Associated With Logged Total Costs of Providing Select Public Health Services: Florida Department of Health, the National Association of County and City Health Officials, and the Area Resource File, Florida, 2008 and 2010 Parameter Estimates Scale of Services Offered Servicesa Services squaredb
Communicable Disease Surveillance and Investigation (n = 111) 0.48** –0.19**
Chronic Disease Prevention (n = 115) 0.23 –0.09
Food Hygiene (n = 115)
On-site Sewage Treatment and Disposal (n = 111)
Vital Records (n = 111)
1.95** –0.31**
3.50** –0.013**
0.074** –0.00034** –1.35
Scope of services offered Clinical preventive services, %
–0.76
0.95
0.59
–0.07
Medical care services, %
–0.67
0.32
–0.35
–0.15
0.12
Specialty services, %
0.13
–0.30
–0.22
0.22
–0.25
Population-based services, %
1.46
1.53*
0.14
–0.31
–0.78
Regulatory and licensing, %
0.87
0.34
0.52
0.10
0.67
Environmental health, % Public health agency characteristics
0.04
–0.82
–0.31
–0.10
0.46
Revenue from local sources, %
–3.47
–1.56
2.39
–0.20
–0.15
Revenue from state sources, %
–3.23
2.10
–0.47
–0.79
–2.70
Revenue from federal sources, %
–3.16
2.88
1.55
–1.37*
–4.91
Revenue from clinical services, %
–3.90*
1.12
–0.46
–0.79
–0.74
1000 persons per square mile
–0.02
0.88**
0.26
0.08
–0.05
Population non-White, % Persons with college education, %
–0.01 11.31*
0.01 2.27
0.02* 4.79
0.00 1.78
–0.01 8.34*
Persons aged ‡ 65 y, %
0.85
–0.24
2.39
–1.02
–0.96
Income per capita (log)
1.26
1.20
0.30
0.10
–0.68
Percent uninsured, %
0.00
0.02
0.02
–0.02
0.04
Year 2008c
0.22
0.72**
0.14
0.16
–0.04
Population and community characteristics
Model Constant Adjusted R2
–2.14
–5.62
3.39
11.71**
18.10*
0.68
0.56
0.84
0.86
0.71
a
Services were expressed in terms of 1000 units. Services squared were expressed in terms of 1 000 000 (1000 x 1000) units provided. Reference year 2010. *P < .05; **P < .01. b c
Supplement 2, 2015, Vol 105, No. S2 | American Journal of Public Health
Bernet and Singh | Peer Reviewed | Research and Practice | S263
RESEARCH AND PRACTICE
a
b 700
1 200 000
600
1 000 000
500
Cost, $
Cost, $
800 000 600 000
400 300
400 000
200
200 000
100
0
0
5 000
10 000
15 000
20 000
25 000
0
0
5 000
10 000
15 000
20 000
25 000
Number of Services
Number of Services
FIGURE 1—Expense and number of services in local health departments for communicable disease surveillance and investigation by (a) total cost and (b) cost per service: Florida Department of Health, the National Association of County and City Health Officials, and the Area Resource File, Florida, 2008 and 2010.
are similar, subtle differences might point to differences in production processes. For example, the cost per service curves for communicable disease surveillance and investigation bends sharply at a relatively low volume, showing fairly flat cost per service beyond that threshold level. This indicated that LHDs might benefit from economies of scale up to the threshold level, but that they might not realize meaningful gains beyond. By contrast, the more gradual slope for per service costs of onsite sewage treatment and disposal indicates that scale economies appeared to be present across the entire range of service volume. Economies of scale can also be thought of in per unit terms, so we duplicated study regressions to estimate cost per service. These alternate models again found scale economies for communicable disease surveillance and investigation, on-site sewage inspection, vital records, and chronic disease prevention services, all of which had negative coefficients on volume, which indicated lower cost per unit at higher volumes. These per unit cost models had the advantage of being more comparable across services, because they were less subject to the sheer impact of population size that was demonstrated in earlier studies.4---6 We also found another perspective on how efficiency considers cost per client. We re-ran study models to estimate total costs based on number of clients rather than service volume. All services demonstrated economies of scale
in per client costs, which was likely caused by the fairly uniform services-per-client intensity across all LHDs. In addition to the fine-grained service classifications noted earlier, Florida has rigorous procedures in place to ensure consistent coding and counting of services, costs, and clients. Looking at scope effects, we found no evidence of meaningful economies of scope for any of the 5 public health activities examined in this study (Table 3). With 1 exception, all coefficients on indicators of the scope of services provided were not significant. Similarly, public health agency, population, and community characteristics were largely unrelated to the cost of service provision (Table 3).
DISCUSSION LHDs varied considerably in size, range of services offered, and the demographic characteristics of the population served. The cost of providing public health services also varied widely. Our findings regarding the cost of delivering services, and the agency, population, and community characteristics that influenced those costs, provided valuable information for agency decisions regarding the scale and scope of public health operations. Our study was among the first to systematically search for economies of scale and scope in the delivery of public health activities. Our findings indicated that economies of scale arose
S264 | Research and Practice | Peer Reviewed | Bernet and Singh
in the production of many core public health activities. This was hardly surprising. Many public health services require a significant level of fixed capacity or staffing, which results in high fixed costs. The recordkeeping infrastructure behind vital records activities, for instance, requires considerable investment in information technologies. Spreading these fixed costs across more services could help health districts reduce per service costs. The presence of economies of scale in the delivery of public health services had important implications for the organization of public health at the local level. Increasing the number of services provided in certain categories might allow LHDs to reduce the cost per unit of service, thus freeing up resources for use in other areas. Evidence of economies of scale could thus inform discussions of local health district consolidation, regionalization, and the provision of shared services. In situations in which an LHD simply does not need the number of services necessary to achieve economies of scale, specialization might help. If 1 LHD provides the service to a number of other LHDs, all would benefit from lower per unit costs. However, focusing attention on a larger production process for 1 service takes attention away from other services, for which local needs are also too low to achieve economies. Rather than each LHD trying to do every service alone, a strategy of specialization and sharing would help LHDs get more for the same cost.
American Journal of Public Health | Supplement 2, 2015, Vol 105, No. S2
RESEARCH AND PRACTICE
a
b 800 000
250
700 000
200
500 000
Cost, $
Cost, $
600 000
400 000
100
300 000 200 000
50
100 000 0
150
0
0
1 000 2 000 3 000 4 000 5 000 6 000 7 000 8 000 9 000 10 000
0
1 000 2 000 3 000 4 000 5 000 6 000 7 000 8 000 9 000 10 000
Number of Services
Number of Services
FIGURE 2—Expense and number of services in local health departments for food hygiene by (a) total cost and (b) cost per service: Florida Department of Health, the National Association of County and City Health Officials, and the Area Resource File, Florida, 2008 and 2010.
An individual LHD might, for example, provide food hygiene services for the entire region, but discontinue the direct provision of on-site sewage inspection in favor of contracted services from another LHD in that region. This broader perspective allows LHDs, as a group, to achieve the economies of scale that none could reach individually. At that individual level, however, cost savings are just 1 factor in strategic decisions that must also consider
strong desires for local governance, potential legal barriers to sharing services, management practices and staff training needed to enable large-scale production, special circumstances of small and rural LHDs, concerns that savings in 1 area must be returned rather than invested in other services,22 and the degree of similarity of adjacent LHD populations that could moderate efficiency levels actually realized. If these factors align, LHDs have a chance to provide more
a
b 2 000 000
450.00
1 800 000
400.00
1 600 000
350.00
1 400 000
300.00
1 200 000
Cost, $
Cost, $
services, effectively improving population health for no additional cost. Findings of economies of scale, however, also pose challenges, especially for smaller LHDs. Higher costs per unit at lower volumes mean that smaller LHDs might not be able to engage in the same level of public health activities, even if their budgets and staffing per capita are equal to those of larger LHDs. Staffing decisions can be challenging at low
1 000 000 800 000
200.00 150.00
600 000
100.00
400 000
50.00
200 000 0
250.00
0.00 0
5 000
10 000
15 000
20 000
Number of Services
25 000
0
5 000
10 000
15 000
20 000
Number of Services
FIGURE 3—Expense and number of services in local health departments for on-site sewage treatment and disposal by (a) total cost and (b) cost per service: Florida Department of Health, the National Association of County and City Health Officials, and the Area Resource File, Florida, 2008 and 2010.
Supplement 2, 2015, Vol 105, No. S2 | American Journal of Public Health
Bernet and Singh | Peer Reviewed | Research and Practice | S265
RESEARCH AND PRACTICE
volumes. Highly specialized workers are often more efficient, but that very specialization limits the functions they can perform. This can leave LHDs with a choice between paying for highly trained workers who are overqualified (and overpaid) for many of the tasks to which they are assigned or using lower qualified workers who may be less efficient. Our findings indicated that small LHDs might thus need to rely on contracted services or partnerships with larger LHDs when attempting to reduce the cost of service provision. Unfortunately, the potential cost savings depend largely on the rate at which services could be subcontracted, and smaller LHDs might lack the bargaining power to negotiate favorable rates.26,27 The difficulties experienced by smaller LHDs might also explain the broader variation in cost per unit at lower volumes that we found (Figures 1---3), making managing smaller LHDs even more of a challenge. By contrast, the breadth of services provided by a LHD was not associated with meaningful variation in the cost of service provision. These findings suggested that, at least for the activities we examined in this study, there were neither economies nor diseconomies of scope. This was an encouraging finding in a public health setting in which many LHDs were expected to provide at least some minimum package of core services. Because the scope of services did not appear to affect the costs of performing individual public health activities, decisions regarding the range of public health services provided might be based on factors other than financial ones and instead follow strategic considerations because of legal requirements.
Limitations Any generalization of our findings must acknowledge the limitations of our study, which we performed using data from LHDs in only 1 state, Florida, and only for 2 years, 2008 and 2010. Many LHDs experienced resource cuts and higher service demand during this time,10 and future research is needed to understand the factors that drive LHDs’ costs in times of stable or growing budgets. Moreover, our cross-sectional study design did not allow inferences regarding the causality of the relationships presented. Longitudinal studies might be better able to identify possible endogeneity issues, wherein LHDs could self-select
lower volume because of higher costs. Although economic theory predicts that greater volume of services will result in lower costs, it is theoretically possible that LHDs may self-select lower volume because of higher costs. In addition, we only studied 5 services, so these services might not be generalizable to others. Exporting findings to other states would require further investigation into other LHD characteristics, such as executive training, board of health powers, and recognition that Florida is a centralized state. Finally, replicating total cost, per unit cost, and per capita models with other services could lead to a deeper understanding of the role of service volume, population size, and breadth of service coverage in the cost of delivering public health services.
Conclusions 1,2
We continued a line of research that examined the cost of providing public health services. The strength of our study was that the models presented were able to explain a significant portion of the variation in costs. This made these models a useful tool for both public health practitioners and researchers. Resource requirements could be estimated with greater precision using such models to better inform policymakers about funding requirements. An understanding of economies of scale is also relevant to discussions about consolidating the production of services and the creating of shared programs across LHDs. Funds saved as a result could be spent on other public health activities, resulting in a net improvement in the health of a population at no additional cost. j
Acknowledgments This project was partly funded by a grant from the Robert Wood Johnson Foundation (grant N016299).
Human Participant Protection This study was approved by the University of Michigan. No human participants were involved.
References 1. Committee on Public Health Strategies to Improve Health; Board on Population Health and Public Health Practices; Institute of Medicine. For the Public’s Health: Investing in a Healthier Future. Available at: https:// download.nap.edu/catalog.php?record_id=13268. Accessed March 3, 2013. 2. Leider JP, Resnick B, Kass N, et al. Budget- and priority-setting criteria at state health agencies in times of austerity: a mixed methods study. Am J Public Health. 2014;104(6):1092---1099. 3. Willard R, Shah GH, Leep C, Ku L. Impact of the 2008-2010 economic recession on local health departments. J Public Health Manag Pract. 2012;18(2):106---114. 4. Santerre RE. Jurisdiction size and local public health spending. Health Serv Res. 2009;44(6):2148---2166. 5. Mays GP, McHugh MC, Shim K, et al. Institutional and economic determinants of public health system performance. Am J Public Health. 2006;96(3):523---531. 6. Erwin PC. The performance of local health departments: a review of the literature. J Public Health Manag Pract. 2008;14(2):E9---E18. 7. Ingram RC, Bernet PM, Costich JF. Public health services and systems research: current state of finance research. J Public Health Manag Pract. 2012;18(6): 515---519. 8. Singh SR. Efficiency in public health service delivery: an analysis of clinical health services provided by local health departments in Florida. Frontiers Public Health Serv Sys Res. 2013;2(7):Article 3. 9. Honoré PA, Amy BW. Public health finance: fundamental theories, concepts, and definitions. J Public Health Manag Pract. 2007;13(2):89---92. 10. Wooldridge JM. Introductory Econometrics: A Modern Approach. 5th ed. Mason, OH: South-Western Cengage Learning; 2012.
About the Authors Patrick M. Bernet is with the College of Business, Florida Atlantic University, Boca Raton. Simone Singh is with the School of Public Health, University of Michigan, Ann Arbor. Correspondence should be sent to Patrick M. Bernet, Florida Atlantic University, College of Business, Department of Health Administration, 127 Fleming West, 777 Glades Road, Boca Raton, FL 33431 (e-mail: PBernet@ fau.edu). Reprints can be ordered at http://www.ajph.org by clicking the “Reprints” link. This article was accepted September 18, 2014.
Contributors This project was jointly conceptualized by P. M. Bernet and S. Singh. P. M. Bernet performed the background research, organized data, and led the writing of the article. S. Singh contributed to the background research,
S266 | Research and Practice | Peer Reviewed | Bernet and Singh
performed statistical analysis, and assisted in writing the technical parts of the article.
11. Getzen TE. Health Economics and Financing. 5th ed. New York, NY: Wiley; 2012. 12. Institute of Medicine. The Future of the Public’s Health in the 21st Century. Washington, DC: The National Academies Press; 2002. Available at: http://www.iom. edu/Reports/2002/The-Future-of-the-Publics-Healthin-the-21st-Century.aspx. Accessed December 12, 2012. 13. Frieden TR. A framework for public health action, the health impact pyramid. Am J Public Health. 2010; 100(4):590---595. 14. Bridges CB, Thompson W, Meltzer MI, et al. Efficacy and cost benefit of influenza vaccination of healthy working adults: a randomized controlled trial. JAMA. 2000;284:1655---1663. 15. Amico PR, Chilingerian JA, van Hasselt M. Community health center efficiency: the role of grant
American Journal of Public Health | Supplement 2, 2015, Vol 105, No. S2
RESEARCH AND PRACTICE
revenues in health center efficiency. Health Serv Res. 2014;49:666---682. 16. Keers JC, Groen H, Sluiter WJ, Bouma J, Links TP. Cost and benefits of a multidisciplinary intensive diabetes education programme. J Eval Clin Pract. 2005;11(3): 293---303. 17. Chesson HW, Greenberg JB, Hennessy M. The cost effectiveness of the WINGS intervention: a program to prevent HIV and sexually transmitted diseases among high risk urban women. BMC Infect Dis. 2002;2:24. 18. Leep CJ, Shah GH. NACCHO’s national profile of local health departments study: the premier source of data on local health departments for surveillance, research, and policymaking. J Public Health Manag Pract. 2012;18(2):186---189. 19. Division of Environmental Health. Environmental Health Coding Pamphlet. DHP-50-21 (V2). Tallahassee, FL: Florida Department of Health; 2011. 20. Department of Health. Personal Health Coding Pamphlet. DHP-50-20. Tallahassee, FL: Florida Department of Health; 2012. 21. Mays GP, Smith S. Geographic variation in public health spending: correlates and consequences. Health Serv Res. 2009;44(5 pt 2):1796---1817. 22. Bernet PM. Local public health agency funding: money begets money. J Public Health Manag Pract. 2007;13(2):188---193. 23. Erwin PC, Greene SB, Mays GP, Ricketts TC, Davis MV. The association of changes in local health department resources with changes in state-level health outcomes. Am J Public Health. 2011;101(4):609---615. 24. Reschovsky A, Zahner S, Enami K, Goitom S. Forecasting the impact of the economic recession on the financing of local public health departments in Wisconsin. Presented at: Public Health Practice Based Research Networks Meeting [webinar]: August 16, 2012. 25. Mukherjee K, Santerre RE, Zhang NJ. Explaining the efficiency of local health departments in the US: an exploratory analysis. Health Care Manag Sci. 2010; 13(4):378---387. 26. Gollust SE, Jacobson PD. Privatization of public services: organizational reform efforts in public education and public health. Am J Public Health. 2006;96(10): 1733---1739. 27. Keane C, Marx J, Ricci E. Perceived outcomes of public health privatization: a national survey of local health department directors. Milbank Q. 2001;79(1): 115---137.
Supplement 2, 2015, Vol 105, No. S2 | American Journal of Public Health
Bernet and Singh | Peer Reviewed | Research and Practice | S267
